[
https://issues.apache.org/jira/browse/CASSANDRA-13673?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Blake Eggleston updated CASSANDRA-13673:
----------------------------------------
Reviewer: Marcus Eriksson
Status: Patch Available (was: Open)
If the repair executor was shutdown before the commit message is sent, none of
the replicas will receive commit messages, so none of IR sessions will
complete. This is because the repair complete callback shuts down the executor
in question. For some reason I made the message sending happen in another
thread, but since MessagingService.send* just puts stuff on a queue (or worst
case starts a few messaging threads), this seems like overkill. Making the
sendMessage stuff happen synchronously fixes the issue.
[trunk|https://github.com/bdeggleston/cassandra/tree/13673]
[utests|https://circleci.com/gh/bdeggleston/cassandra/63]
> Incremental repair coordinator sometimes doesn't send commit messages
> ---------------------------------------------------------------------
>
> Key: CASSANDRA-13673
> URL: https://issues.apache.org/jira/browse/CASSANDRA-13673
> Project: Cassandra
> Issue Type: Bug
> Reporter: Blake Eggleston
> Assignee: Blake Eggleston
> Fix For: 4.0
>
>
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
